Risk
2/8/2013
04:49 PM
Connect Directly
LinkedIn
Twitter
Google+
RSS
E-Mail
50%
50%

Facebook Login Bug: Lessons Learned

Service interruption at dozens of prominent websites including CNN and Hulu reminds that third-party code integration carries risks.

Soltani likens the situation to using giving a speech using the Mad Libs format. You allow a third-party to supply you with words periodically and hope that those words don't alter what you meant to say or embarrass you.

"As a result of a Facebook bug, entire domains were redirected," said Soltani. "That's an incredible amount of power to have."

Facebook presumably isn't going to be introducing similar bugs on a frequent basis. But the online ad industry deals with code trust problems every day.

Websites allow ads to be injected into their webpage code in the same way that many sites allow Facebook Login code to be injected. And in the ad business, the tenuous chain of connections across multiple domains and servers makes trust much more difficult. Cyber criminals routinely attempt to subvert ad network security, through hacking and fraudulent business practices, to distribute malicious code.

Soltani said that combined with other malicious activity such as DNS hijacking, reliance on third-party injected code can be even more of a problem.

There's also the issue of performance. Until December 2009, Google Analytics -- incorporated into many websites -- wasn't asynchronous, meaning that calls to Google's servers made webpages load more slowly. Other widget code inserted into webpages often has a similar effect.

In a blog post last October, Aarron Walter, director of user experience at email marketing service MailChimp, argued that social logins aren't worth it, a position others have taken before him. The social login buttons offered by Facebook and Twitter, he said, put security in someone else's hands and present the risk of brand damage if problems occur.

Moreover, the rationale for implementing social login buttons -- user convenience and ease of use -- might not be justified. Walter found that a 66% decrease in failed logins and a 42% decrease in password resets at Mail Chimp last year were the result of user interface design improvements and better programming, or "better error handling and copywriting" as he put it. Social logins contributed to only 3.4% of the drop in login problems.

Further complicating the question of whether to use a social login is the finding that many companies have implemented Facebook Login ineffectively. Research published by marketing consultancy Social Labs in September 2012 indicates that among the top 500 online retailers, only 30 have implemented a social login -- 77% of these being Facebook Login -- and a fair number of these have done so incorrectly. For example, 30% of the 30 online retailers using a social login require a website password in tandem with a visitor's social username and password, and 15 of these online retailers don't offer a social login during the checkout process, making for an inconsistent user experience. Finally, none of online retailers surveyed have implemented Facebook's login persistence, which automatically logs users in when they return to the website.

Despite these issues, Sociable Labs expects adoption of social logins to increase over time.

Previous
2 of 2
Next
Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
lgarey@techweb.com
50%
50%
lgarey@techweb.com,
User Rank: Apprentice
2/11/2013 | 4:04:36 PM
re: Facebook Login Bug: Lessons Learned
The Mad Libs example is excellent. Is there any indication that this was not just a bug but a hack? Lorna Garey, IW Reports
Register for Dark Reading Newsletters
White Papers
Cartoon
Current Issue
Flash Poll
Video
Slideshows
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2014-4774
Published: 2015-05-25
Cross-site request forgery (CSRF) vulnerability in the login page in IBM License Metric Tool 9 before 9.1.0.2 and Endpoint Manager for Software Use Analysis 9 before 9.1.0.2 allows remote attackers to hijack the authentication of arbitrary users via vectors involving a FRAME element.

CVE-2014-4778
Published: 2015-05-25
IBM License Metric Tool 9 before 9.1.0.2 and Endpoint Manager for Software Use Analysis 9 before 9.1.0.2 do not send an X-Frame-Options HTTP header in response to requests for the login page, which allows remote attackers to conduct clickjacking attacks via vectors involving a FRAME element.

CVE-2014-6190
Published: 2015-05-25
The log viewer in IBM Workload Deployer 3.1 before 3.1.0.7 allows remote attackers to obtain sensitive information via a direct request for the URL of a log document.

CVE-2014-6192
Published: 2015-05-25
Cross-site scripting (XSS) vulnerability in IBM Curam Social Program Management 6.0 SP2 before EP26, 6.0.4 before 6.0.4.5 iFix10, 6.0.5 before 6.0.5.6, and 6.0.5.5a before 6.0.5.8 allows remote authenticated users to inject arbitrary web script or HTML via a crafted URL.

CVE-2014-8146
Published: 2015-05-25
The resolveImplicitLevels function in common/ubidi.c in the Unicode Bidirectional Algorithm implementation in ICU4C in International Components for Unicode (ICU) before 55.1 does not properly track directionally isolated pieces of text, which allows remote attackers to cause a denial of service (hea...

Dark Reading Radio
Archived Dark Reading Radio
Join security and risk expert John Pironti and Dark Reading Editor-in-Chief Tim Wilson for a live online discussion of the sea-changing shift in security strategy and the many ways it is affecting IT and business.